Adaptive Neuro Fuzzy Inference System based Optical Character Recognition

Optical character recognition (OCR) is becoming a powerful tool in the field of Character Recognition, now a days. In the existing globalized environment, OCR can play a vital role in different application fields. Basically, OCR technique converts images into editable format. This technique converts images in the form of documents such as we can edit, modify and store data more safely for long time. We need to train our classifier in case we are considering using data mining techniques for such purposes. There are several established generic classification techniques that can be used together with feature extraction mechanisms but it is important to know which of them do better under which circumstances. This paper describes the application of adaptive neuro-fuzzy inference system (ANFIS) model for classification of machine printed characters. Decision making was performed in two stages: feature extraction using the zoning technique and the ANFIS trained with the backpropagation gradient descent method in combination with the least squares method. To improve diagnostic accuracy, the sixth ANFIS classifier (combining ANFIS) was trained using the outputs of the five ANFIS classifiers as input data. The proposed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach. Some conclusions concerning the saliency of features on classification were obtained through analysis of the ANFIS. The performance of the ANFIS model was evaluated in terms of training performance and classification accuracies and the results confirmed that the proposed ANFIS model has potential in classifying the machine printed characters.